Chapter 4 - Modelling
Chapter 4 - Modelling
Chapter 4 - Modelling
3 equilibrium equations
𝜕𝜎𝑥𝑥 𝜕𝜎𝑥𝑦 𝜕𝜎𝑥𝑧
+ + + 𝜌𝑓𝑥 = 𝜌𝑎𝑥
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕𝜎𝑦𝑥 𝜕𝜎𝑦𝑦 𝜕𝜎𝑦𝑧
+ + + 𝜌𝑓𝑦 = 𝜌𝑎𝑦
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕𝜎𝑧𝑥 𝜕𝜎𝑧𝑦 𝜕𝜎𝑧𝑧
+ + + 𝜌𝑓𝑧 = 𝜌𝑎𝑧
𝜕𝑥 𝜕𝑦 𝜕𝑧
6 strain-displacement equations
𝜕𝑢 𝜕𝑣 𝜕𝑤
𝜀𝑥𝑥 = , 𝜀𝑦𝑦 = , 𝜀𝑧𝑧 =
𝜕𝑥 𝜕𝑦 𝜕𝑧
1 𝜕𝑣 𝜕𝑢 1 𝜕𝑤 𝜕𝑣 1 𝜕𝑢 𝜕𝑤
𝜀𝑥𝑦 = ( + ) , 𝜀𝑦𝑧 = ( + ) , 𝜀𝑧𝑥 = ( + )
2 𝜕𝑥 𝜕𝑦 2 𝜕𝑦 𝜕𝑧 2 𝜕𝑧 𝜕𝑥
6 stress-strain equations
𝐸 𝐸
𝜎𝑥𝑥 = [(1 − 𝜈)𝜀𝑥𝑥 + 𝜈(𝜀𝑦𝑦 + 𝜀𝑧𝑧 )] 𝜎𝑥𝑦 = 𝜀
(1 + 𝜈)(1 − 2𝜈) (1 + 𝜈) 𝑥𝑦
𝐸 𝐸
𝜎𝑦𝑦 = [(1 − 𝜈)𝜀𝑦𝑦 + 𝜈(𝜀𝑧𝑧 + 𝜀𝑥𝑥 )] 𝜎𝑦𝑧 = 𝜀
(1 + 𝜈)(1 − 2𝜈) (1 + 𝜈) 𝑦𝑧
𝐸 𝐸
𝜎𝑧𝑧 = [(1 − 𝜈)𝜀𝑧𝑧 + 𝜈(𝜀𝑥𝑥 + 𝜀𝑦𝑦 )] 𝜎𝑧𝑥 = 𝜀
(1 + 𝜈)(1 − 2𝜈) (1 + 𝜈) 𝑧𝑥
1
MECH3361/9361 Semester 2, 2016
Top view
z
y Cutting
zone A
x
Cutting
tool
Here’s what we know about the cutting tool. Firstly, since the cutting tool is clamped onto the
tool post or the square turret of the lathe, the clamped end is not able to move. Secondly,
some cutting force is applied in the cutting zone. Thus, the problem can be translated into the
mechanics model shown to the bottom right of Fig. 4.1.
Displacement BCs
(e.g. at 𝑥 = 0, 𝑢 = 𝑣 = 𝑤 = 0)
Stress BCs
(e.g. at the tool-workpiece interaction zone, 𝜎𝑛𝑥 = 𝐹𝑥 , 𝜎𝑛𝑦 = 𝐹𝑦 , 𝜎𝑛𝑧 = 𝐹𝑧 )
2
MECH3361/9361 Semester 2, 2016
Stress B.C.
Displacement
B.C.
Mixed
B.C.
Fig. 4.2 Boundary conditions
𝑢 = 𝑢 ′ , 𝑣 = 𝑣 ′ , 𝑤 = 𝑤′
′′ ′′ ′′
Similarly, if the stresses at a surface point (𝑥”, 𝑦”, 𝑧”) are known as 𝜎𝑛𝑥 , 𝜎𝑛𝑦 , 𝜎𝑛𝑧 , the correct
solution to stress functions 𝜎𝑥𝑥 , 𝜎𝑦𝑦 , 𝜎𝑧𝑧 , 𝜎𝑥𝑦 , 𝜎𝑦𝑧 , 𝜎𝑧𝑥 must satisfy the stress B.C. as:
′′
𝜎𝑛𝑥 = 𝑙𝜎𝑥𝑥 + 𝑚𝜎𝑦𝑥 + 𝑛𝜎𝑧𝑥
′′
{𝜎𝑛𝑦 = 𝑙𝜎𝑥𝑦 + 𝑚𝜎𝑦𝑦 + 𝑛𝜎𝑧𝑦
′′
𝜎𝑛𝑧 = 𝑙𝜎𝑥𝑧 + 𝑚𝜎𝑦𝑧 + 𝑛𝜎𝑧𝑧
BCs are an important mathematical requirement in solving field problems. For mechanics
modelling, we use a number of differential equations to model the underlying physics. Since
differential equations only provide relationships from one point to the next, absolute values
(for displacement, etc.) cannot be calculated unless appropriate BCs are applied.
Example 4.1
A prismatic squared beam (𝑎×𝑎×𝐿) is clamped on its sides. A heavy cubic object (𝑎×𝑎×𝑎) is
placed on top of the beam. The density of this object is 𝜌. Describe the BCs of the beam.
3
MECH3361/9361 Semester 2, 2016
Solution
For a complete answer, we must define both the surfaces and the associated BC
mathematically.
−𝑎 ≤ 𝑥 ≤ 0, 𝑦 = 0, −𝑎 ≤ 𝑧 ≤ 0
and EFGH:
−𝑎 ≤ 𝑥 ≤ 0, 𝑦 = 𝐿, −𝑎 ≤ 𝑧 ≤ 0
𝑢=𝑣=𝑤=0
𝑥 = 0, 0 ≤ 𝑦 ≤ 𝐿, −𝑎 ≤ 𝑧 ≤ 0
and DCGH:
𝑥 = −𝑎, 0 ≤ 𝑦 ≤ 𝐿, −𝑎 ≤ 𝑧 ≤ 0
there are no stresses (free x-surfaces), so the stress BC at these surfaces is:
−𝑎 ≤ 𝑥 ≤ 0, 0 ≤ 𝑦 ≤ 𝑏, 𝑧 = 0
KEHL:
−𝑎 ≤ 𝑥 ≤ 0, (𝑎 + 𝑏) ≤ 𝑦 ≤ 𝐿, 𝑧 = 0
and BFGC:
−𝑎 ≤ 𝑥 ≤ 0, 0 ≤ 𝑦 ≤ 𝐿, 𝑧 = −𝑎
there are no stresses (free z-surfaces), so the stress BC at these surfaces is:
−𝑎 ≤ 𝑥 ≤ 0, 𝑏 ≤ 𝑦 ≤ (𝑎 + 𝑏), 𝑧 = 0
4
MECH3361/9361 Semester 2, 2016
Sliding block 1
P
b
x
Slide way
b a a
2
On the top and bottom surfaces, there is a displacement BC in the normal (y) direction, but
there is also a friction stress in the tangential (x) direction. Therefore, we can define the BCs
as follows:
To denote symmetry, we use the fact that no displacement can occur across the axis, since
this would violate the compatibility of strains. For the through-hole plate below, we can
reduce the problem domain because of the symmetry about both the x and y axes. Left-right
symmetry about the (vertical) y-axis implies that at 𝑥 = 0, 𝑢 = 0. Conversely, there is top-
bottom symmetry about the (horizontal) x-axis, so we know that at 𝑦 = 0, 𝑣 = 0. These
displacement restrictions are illustrated using the fixed links as shown for the quarter-plate.
5
MECH3361/9361 Semester 2, 2016
y
𝜎 = 1 MPa
C
I H
ux=0
A a
2m
B
x
uy=0
Mild Steel
F G
2m
4.3.2. Definition
Many engineering structures are made of thin plate with the loads applied in the plane, e.g. a
movable white board in the classroom.
F F
t h,b
h
y
mg
x
z b
Fig. 4.4 Example of plane stress
In such a problem, 𝑡 ≤ ℎ, 𝑏 and all the loads are applied in-plane as shown in Fig. 4.4. On the
two stress-free surfaces of the plate (front and back):
Thus, the non-zero stress components are 𝜎𝑥𝑥 , 𝜎𝑦𝑦 , and 𝜎𝑥𝑦 . These stress components are
independent of coordinate 𝑧, i.e.
6
MECH3361/9361 Semester 2, 2016
𝜎𝑥𝑥 = 𝜎𝑥𝑥 (𝑥, 𝑦), 𝜎𝑦𝑦 = 𝜎𝑦𝑦 (𝑥, 𝑦), 𝜎𝑥𝑦 = 𝜎𝑥𝑦 (𝑥, 𝑦)
or in other words, all of the non-vanishing stresses are in the xy-plane only. We therefore call
these “plane stress” problems.
From generalised Hooke’s law, we can see that all strains components are also independent
of 𝑧:
𝜀𝑥𝑥 = 𝜀𝑥𝑥 (𝑥, 𝑦), 𝜀𝑦𝑦 = 𝜀𝑦𝑦 (𝑥, 𝑦), 𝜀𝑥𝑦 = 𝜀𝑥𝑦 (𝑥, 𝑦)
𝜀𝑧𝑥 = 𝜀𝑧𝑦 = 0
1 𝜈
𝜀𝑧𝑧 = [𝜎𝑧𝑧 − 𝜈(𝜎𝑥𝑥 + 𝜎𝑦𝑦 )] = − (𝜎𝑥𝑥 + 𝜎𝑦𝑦 )
𝐸 𝐸
Note that 𝜀𝑧𝑧 ≠ 0, but it is not an independent variable (it can be derived from 𝜎𝑥𝑥 and 𝜎𝑦𝑦 ).
𝜕𝑤 𝜈
= 𝜀𝑧𝑧 = − (𝜎𝑥𝑥 + 𝜎𝑦𝑦 )
𝜕𝑧 𝐸
Stress: 𝜎𝑥𝑥 = 𝜎𝑥𝑥 (𝑥, 𝑦), 𝜎𝑦𝑦 = 𝜎𝑦𝑦 (𝑥, 𝑦), 𝜎𝑥𝑦 = 𝜎𝑥𝑦 (𝑥, 𝑦)
Strain: 𝜀𝑥𝑥 = 𝜀𝑥𝑥 (𝑥, 𝑦), 𝜀𝑦𝑦 = 𝜀𝑦𝑦 (𝑥, 𝑦), 𝜀𝑥𝑦 = 𝜀𝑥𝑦 (𝑥, 𝑦)
Displacements: 𝑢, 𝑣
𝜕𝜎𝑥𝑥 𝜕𝜎𝑥𝑦
+ + 𝑏𝑥 = 𝜌𝑎𝑥
𝜕𝑥 𝜕𝑦
Equilibrium: {𝜕𝜎 𝜕𝜎𝑦𝑦
𝑦𝑥
+ + 𝑏𝑦 = 𝜌𝑎𝑦
𝜕𝑥 𝜕𝑦
𝜕𝑢
𝜀𝑥𝑥 = 𝜕𝑥
𝜕𝑣
Strain-displacement: 𝜀𝑦𝑦 = 𝜕𝑦
1 𝜕𝑢 𝜕𝑣
{𝜀𝑥𝑦 = 2 (𝜕𝑦 + 𝜕𝑥)
𝐸𝜀𝑥𝑥 = 𝜎𝑥𝑥 − 𝜈𝜎𝑦𝑦
Stress-strain relationship: {𝐸𝜀𝑦𝑦 = 𝜎𝑦𝑦 − 𝜈𝜎𝑥𝑥
𝐸𝜀𝑥𝑦 = (1 + 𝜈)𝜎𝑥𝑦
7
MECH3361/9361 Semester 2, 2016
In this case, the strains in the z-direction (longitudinal) are effectively zero (since that
dimension is much longer than the other two):
which implies that the displacement in the z-direction (𝑤) vanishes. Also, since the loading is
uniform, the stresses, strains, and displacements in any two cross-sections parallel to the
xy-plane must be identical. This means that once again, all the non-vanishing stresses,
strains, and displacements are functions of x and y, and all lie within the xy-plane, i.e.
𝜀𝑥𝑥 = 𝜀𝑥𝑥 (𝑥, 𝑦), 𝜀𝑦𝑦 = 𝜀𝑦𝑦 (𝑥, 𝑦), 𝜀𝑥𝑦 = 𝜀𝑥𝑦 (𝑥, 𝑦)
1 1+𝜈
𝜀𝑥𝑥 = [𝜎 − 𝜈(𝜎𝑦𝑦 + 𝜎𝑧𝑧 )] 𝜀𝑥𝑦 = 𝜎𝑥𝑦
𝐸 𝑥𝑥 𝐸
1 1+𝜈
𝜀𝑦𝑦 = [𝜎 − 𝜈(𝜎𝑧𝑧 + 𝜎𝑥𝑥 )] 𝜀𝑦𝑧 = 𝜎𝑦𝑧
𝐸 𝑦𝑦 𝐸
1 1+𝜈
𝜀𝑧𝑧 = [𝜎 − 𝜈(𝜎𝑥𝑥 + 𝜎𝑦𝑦 )] 𝜀𝑧𝑥 = 𝜎𝑧𝑥
𝐸 𝑧𝑧 𝐸
8
MECH3361/9361 Semester 2, 2016
1
𝜀𝑥𝑥 = [𝜎 − 𝜈(𝜎𝑦𝑦 + 𝜎𝑧𝑧 )]
𝐸 𝑥𝑥
1
= [𝜎𝑥𝑥 − 𝜈 (𝜎𝑦𝑦 + 𝜈(𝜎𝑥𝑥 + 𝜎𝑦𝑦 ))]
𝐸
1
= [(1 − 𝜈 2 )𝜎𝑥𝑥 − 𝜈(1 + 𝜈)𝜎𝑦𝑦 ]
𝐸
𝐸
Multiplying by 1−𝜈 2, we find that:
𝐸 𝜈(1 + 𝜈) 𝜈(1 + 𝜈) 𝜈
𝜀𝑥𝑥 = 𝜎𝑥𝑥 − 𝜎𝑦𝑦 = 𝜎𝑥𝑥 − 𝜎𝑦𝑦 = 𝜎𝑥𝑥 − ( )𝜎
1−𝜈 2 1−𝜈 2 (1 + 𝜈)(1 − 𝜈) 1 − 𝜈 𝑦𝑦
Similarly,
𝐸 𝜈
𝜀𝑦𝑦 = 𝜎𝑦𝑦 − ( )𝜎
1 − 𝜈2 1 − 𝜈 𝑥𝑥
In shear:
𝐸 1 − 𝜈2
𝜎𝑥𝑦 = 𝜀𝑥𝑦 = 𝜎
(1 + 𝜈) 𝐸(1 − 𝜈) 𝑥𝑦
𝐸 1 1+𝜈−𝜈 𝜈
∴ 𝜀𝑥𝑦 = 𝜎𝑥𝑦 = 𝜎𝑥𝑦 = [1 + ]𝜎
1−𝜈 2 1−𝜈 1−𝜈 1 − 𝜈 𝑥𝑦
𝜎𝑧𝑧 = 0
{ 𝑤≠0
𝜈
𝜀𝑧𝑧 = − (𝜎𝑥𝑥 + 𝜎𝑦𝑦 ) ≠ 0
𝐸
𝜀𝑧𝑧 = 0
{ 𝑤=0
𝜎𝑧𝑧 = 𝜈(𝜎𝑥𝑥 + 𝜎𝑦𝑦 ) ≠ 0
9
MECH3361/9361 Semester 2, 2016
𝜈
𝜺𝒛𝒛 𝜀𝑧𝑧 = − (𝜎 + 𝜎𝑦𝑦 ) ≠ 0 0
𝐸 𝑥𝑥
𝐸 𝜈
xx 𝐸𝜀𝑥𝑥 = 𝜎𝑥𝑥 − 𝜈𝜎𝑦𝑦 𝜀𝑥𝑥 = 𝜎𝑥𝑥 − ( )𝜎
1−𝜈 2 1 − 𝜈 𝑦𝑦
𝐸 𝜈
yy 𝐸𝜀𝑦𝑦 = 𝜎𝑦𝑦 − 𝜈𝜎𝑥𝑥 𝜀𝑦𝑦 = 𝜎𝑦𝑦 − ( )𝜎
1 − 𝜈2 1 − 𝜈 𝑥𝑥
𝐸 𝜈
xy 𝐸𝜀𝑥𝑦 = (1 + 𝜈)𝜎𝑥𝑦 𝜀𝑥𝑦 = [1 + ]𝜎
1−𝜈 2 1 − 𝜈 𝑥𝑦
𝐸
E* 𝐸∗ = 𝐸 𝐸∗ =
1 − 𝜈2
𝜈
𝝂∗ 𝜈∗ = 𝜈 𝜈∗ =
1−𝜈
If we have obtained the solution to a plane stress problem, the corresponding plane strain
𝐸 𝜈
solution can be found by simply replacing the elastic constants E and with 1−𝜈2 and 1−𝜈,
respectively.
∗
𝜈
𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 =
1−𝜈
∗
𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 (1 − 𝜈) = 𝜈
10
MECH3361/9361 Semester 2, 2016
∗ ∗
𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 − 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 𝜈=𝜈
∗ ∗
𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 = (1 + 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 )𝜈
∗
𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛
∴𝜈= ∗
1 + 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛
∗
𝐸
𝐸𝑝𝑠𝑡𝑟𝑎𝑖𝑛 =
1 − 𝜈2
∗
𝐸 = 𝐸𝑝𝑠𝑡𝑟𝑎𝑖𝑛 (1 − 𝜈 2 )
∗ 2
∗
𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛
= 𝐸𝑝𝑠𝑡𝑟𝑎𝑖𝑛 [1 − ( ∗ ) ]
1 + 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛
∗ 2 ∗ 2
∗
(1 + 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 ) 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛
= 𝐸𝑝𝑠𝑡𝑟𝑎𝑖𝑛 [ 2 − 2]
∗ ∗
(1 + 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 ) (1 + 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 )
∗
∗
1 + 2𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛
= 𝐸𝑝𝑠𝑡𝑟𝑎𝑖𝑛 [ 2]
∗
(1 + 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 )
𝐸(1+2𝜈)
Therefore, by replacing the E and in the solution of a plane-strain problem by (1+𝜈)2
and
𝜈
, we can obtain the solution to the plane stress problem.
1+𝜈
Example 4.2
A “thin circular plate” of radius R is subjected to a uniform pressure 𝑝1 on its edge. The two
planar plate surfaces are free, as illustrated in the figure below.
(1−𝜈)
The in-plane (xy-plane) displacements in the plate are found to be 𝑢 = −𝑝1 𝑥 and
𝐸
(1−𝜈)
𝑣 = −𝑝1 𝑦. Find the corresponding displacements in a “long circular bar” under a
𝐸
uniform pressure 𝑝2 . Assume that the length of the bar is much longer than its radius R.
11
MECH3361/9361 Semester 2, 2016
Solution
𝐸
From the plane stress solution to a plane strain solution: Using 𝐸 ∗ = 1−𝜈2 to replace 𝐸, 𝜈 ∗ =
𝜈
1−𝜈
to replace 𝜈, and pressure 𝑝2 to replace 𝑝1 :
𝜈
(1 − 𝜈) (1 − 1 − 𝜈) (1 − 𝜈 2 ) − 𝜈(1 + 𝜈)
𝑢 = −𝑝2 𝑥 = −𝑝2 𝐸
𝑥 = −𝑝2 𝑥
𝐸 𝐸
1 − 𝜈2
(1 + 𝜈)(1 − 2𝜈)
= −𝑝2 𝑥
𝐸
Similarly,
(1 + 𝜈)(1 − 2𝜈)
𝑣 = −𝑝2 𝑦
𝐸
y
y
x x
Integration
12
MECH3361/9361 Semester 2, 2016
Example 4.3
For a thin elastic plate under a plane stress scenario, the stress functions for in-plane bending
are known to be 𝜎𝑦𝑦 = 𝜎𝑥𝑦 = 0 and 𝜎𝑥𝑥 = 6𝑎𝑦. For the same plate under in-plane shearing,
the stress functions are 𝜎𝑥𝑥 = 𝜎𝑦𝑦 = 0 and 𝜎𝑥𝑦 = −𝑎. Find the stress in the plate under the
combined loading scenario.
Solution
The stress functions for the two separate loading cases are known and can be illustrated as
below:
-a -a
6ay
x
6ay
y
To get the combined loading, the stress functions can be added directly as per the principle of
superposition. As such:
It is impossible to solve for all of these together. We therefore have to solve one or two
groups first, based on what is known about the problem. Generally speaking, there are four
different methods: the displacement method, stress method, strain method, and mixed
method.
Procedure: The other two sets of unknown variables must first be eliminated from the
equations. This is done by replacing stresses and strains with displacements as follows.
13
MECH3361/9361 Semester 2, 2016
𝜕𝐼1𝜀
(𝜆 + 𝜇) + 𝜇∇2 𝑢 + 𝜌𝑓𝑥 = 𝜌𝑎𝑥
𝜕𝑥
𝜕𝐼1𝜀
(𝜆 + 𝜇) + 𝜇∇2 𝑣 + 𝜌𝑓𝑦 = 𝜌𝑎𝑦
𝜕𝑦
𝜕𝐼 𝜀
(𝜆 + 𝜇) 1 + 𝜇∇2 𝑤 + 𝜌𝑓𝑧 = 𝜌𝑎𝑧
{ 𝜕𝑥
𝜕2 𝜕2 𝜕2 𝜕𝑢 𝜕𝑣 𝜕𝑤
where the Laplace operator ∇2 = 𝜕𝑥 2 + 𝜕𝑦 2 + 𝜕𝑧 2 and 𝐼1𝜀 = 𝜕𝑥 + 𝜕𝑦 + .
𝜕𝑧
After obtaining 𝑢, 𝑣, 𝑤, one can calculate strain by using the strain-displacement equations,
and stress by using Hooke’s law. Note that the solution must satisfy the boundary conditions.
Procedure: Solve for the stress component first, then strains and displacements.
Procedure: Solve for the strain component first, then stresses and displacements.
po r
pi Ri
z
Ro
Given the cylindrical geometry of these vessels, it is convenient to use the cylindrical
coordinate system introduced earlier for analysis. Like the Cartesian system, the cylindrical
system consists of 3 independent coordinates—in this case, (𝑟, 𝜃, 𝑧). The stress tensor for this
system has the following layout.
14
MECH3361/9361 Semester 2, 2016
dr
𝜎𝑟𝑟 𝜎𝑟𝜃 𝜎𝑟𝑧 dz
Equilibrium equations:
Strain-displacement relationships:
𝜕𝑢 1 𝜕𝑣 1 𝜕𝑢 𝑣
𝜀𝑟𝑟 = 𝜀𝑟𝜃 = ( + − )
𝜕𝑟 2 𝜕𝑟 𝑟 𝜕𝜃 𝑟
1 𝜕𝑣 𝑢 1 1 𝜕𝑤 𝜕𝑣
Normal: 𝜀𝜃𝜃 = 𝑟 𝜕𝜃 + 𝑟 Shear: 𝜀𝜃𝑧 = 2 (𝑟 𝜕𝜃 + 𝜕𝑧 )
𝜕𝑤 1 𝜕𝑢 𝜕𝑤
{ 𝜀𝑧𝑧 = { 𝜀𝑧𝑟 = 2 ( 𝜕𝑧 + )
𝜕𝑧 𝜕𝑥
Hooke’s law:
𝐸
𝜎𝑟𝑟 = [(1 − 𝜈)𝜀𝑟𝑟 + 𝜈(𝜀𝜃𝜃 + 𝜀𝑧𝑧 )]
(1 + 𝜈)(1 − 2𝜈)
15
MECH3361/9361 Semester 2, 2016
𝐸
𝜎𝜃𝜃 = [(1 − 𝜈)𝜀𝜃𝜃 + 𝜈(𝜀𝑧𝑧 + 𝜀𝑟𝑟 )]
(1 + 𝜈)(1 − 2𝜈)
𝐸
𝜎𝑧𝑧 = [(1 − 𝜈)𝜀𝑧𝑧 + 𝜈(𝜀𝑟𝑟 + 𝜀𝜃𝜃 )]
(1 + 𝜈)(1 − 2𝜈)
𝐸
𝜎𝑧𝑟 = 𝜀
1 + 𝜈 𝑧𝑟
Step 2: Analysis
The deformation is axisymmetric and under plane strain, so the deformation is independent
of coordinates 𝜃 and 𝑧. Thus, the circumferential and axial displacements, 𝑣 and 𝑤, will
vanish, and the displacements can be expressed as:
𝑢 = 𝑢(𝑟)
𝑣=0
𝑤=0
𝜕𝑢
𝜀𝑟𝑟 =
𝜕𝑟
1 𝜕𝑣 𝑢 1 𝑢 𝑢
𝜀𝜃𝜃 = + = 0+ =
𝑟 𝜕𝜃 𝑟 𝑟 𝑟 𝑟
𝜕𝑤
𝜀𝑧𝑧 = =0
𝜕𝑧
1 𝜕𝑣 1 𝜕𝑢 𝑣 1 𝜕𝑢(𝑟) 1
𝜀𝑟𝜃 = ( + − )= ( )= 0=0
2 𝜕𝑟 𝑟 𝜕𝜃 𝑟 2𝑟 𝜕𝜃 2𝑟
1 1 𝜕𝑤 𝜕𝑣 1 1
𝜀𝜃𝑧 = ( + ) = ( 0 + 0) = 0
2 𝑟 𝜕𝜃 𝜕𝑧 2 𝑟
1 𝜕𝑢 𝜕𝑤 1 𝜕𝑢(𝑟) 1
𝜀𝑧𝑟 = ( + )= ( + 0) = (0 + 0) = 0
2 𝜕𝑧 𝜕𝑥 2 𝜕𝑧 2
𝐸 𝜕𝑢 𝑢
𝜎𝑟𝑟 = [(1 − 𝜈) + 𝜈 ( + 0)]
(1 + 𝜈)(1 − 2𝜈) 𝜕𝑟 𝑟
16
MECH3361/9361 Semester 2, 2016
𝐸 𝜕𝑢 𝑢
= [(1 − 𝜈) +𝜈 ]
(1 + 𝜈)(1 − 2𝜈) 𝜕𝑟 𝑟
𝐸 𝑢 𝜕𝑢
𝜎𝜃𝜃 = [(1 − 𝜈) + 𝜈 (0 + )]
(1 + 𝜈)(1 − 2𝜈) 𝑟 𝜕𝑟
𝐸 𝑢 𝜕𝑢
= [(1 − 𝜈) + 𝜈 ]
(1 + 𝜈)(1 − 2𝜈) 𝑟 𝜕𝑟
𝐸 𝜕𝑢 𝑢
𝜎𝑧𝑧 = [(1 − 𝜈)0 + 𝜈 ( + )]
(1 + 𝜈)(1 − 2𝜈) 𝜕𝑟 𝑟
𝐸𝜈 𝜕𝑢 𝑢
= [ + ]
(1 + 𝜈)(1 − 2𝜈) 𝜕𝑟 𝑟
𝐸 𝐸
𝜎𝑧𝑟 = 𝜀𝑧𝑟 = 0=0
1+𝜈 1+𝜈
For this question, we will assume that body forces are negligible and there is no acceleration.
Therefore, the equations are:
17
MECH3361/9361 Semester 2, 2016
𝜕 𝐸 𝜕𝑢 𝑢
𝐿𝐻𝑆 = { [(1 − 𝜈) + 𝜈 ]}
𝜕𝑟 (1 + 𝜈)(1 − 2𝜈) 𝜕𝑟 𝑟
1 𝐸 𝜕𝑢 𝑢
+ { [(1 − 𝜈) +𝜈 ]
𝑟 (1 + 𝜈)(1 − 2𝜈) 𝜕𝑟 𝑟
𝐸 𝑢 𝜕𝑢
− [(1 − 𝜈) + 𝜈 ]}
(1 + 𝜈)(1 − 2𝜈) 𝑟 𝜕𝑟
𝜕 𝜕𝑢 𝑢 1 𝜕𝑢 𝑢 𝑢 𝜕𝑢
= {(1 − 𝜈) + 𝜈 } + {[(1 − 𝜈) + 𝜈 ] − [(1 − 𝜈) + 𝜈 ]}
𝜕𝑟 𝜕𝑟 𝑟 𝑟 𝜕𝑟 𝑟 𝑟 𝜕𝑟
𝜕 𝜕𝑢 1 1 𝜕𝑢 𝜕𝑢 𝑢 𝑢
= {(1 − 𝜈) + 𝜈 ( ) 𝑢} + { − 2𝜈 − + 2𝜈 }
𝜕𝑟 𝜕𝑟 𝑟 𝑟 𝜕𝑟 𝜕𝑟 𝑟 𝑟
2
𝜕 𝑢 𝑢 𝜈 𝜕𝑢 1 𝜕𝑢 2𝜈 𝜕𝑢 𝑢 𝑢
= {(1 − 𝜈) 2 + (−𝜈 2 + )} + { − − 2 + 2𝜈 2 }
𝜕𝑟 𝑟 𝑟 𝜕𝑟 𝑟 𝜕𝑟 𝑟 𝜕𝑟 𝑟 𝑟
𝜕2𝑢 𝑢 𝜈 𝜕𝑢 1 𝜕𝑢 2𝜈 𝜕𝑢 𝑢 𝑢
= (1 − 𝜈) 2 − 𝜈 2 + + − − 2 + 2𝜈 2
𝜕𝑟 𝑟 𝑟 𝜕𝑟 𝑟 𝜕𝑟 𝑟 𝜕𝑟 𝑟 𝑟
𝜕2𝑢 1 𝜕𝑢 𝑢
= (1 − 𝜈) 2 + (1 − 𝜈) − (1 − 𝜈) 2
𝜕𝑟 𝑟 𝜕𝑟 𝑟
𝜕 2 𝑢 1 𝜕𝑢 𝑢
= 2+ −
𝜕𝑟 𝑟 𝜕𝑟 𝑟 2
=0
Step 6: Solve the above linear and static ordinary differential equation
1
We assume a trial solution 𝑢 = 𝐶1 𝑟 + 𝐶2 𝑟, in which 𝐶1 and 𝐶2 are constants to be determined
by using the BCs.
𝜕𝑢 𝜕 1 1
𝜀𝑟𝑟 = = (𝐶1 𝑟 + 𝐶2 ) = 𝐶1 − 𝐶2 2
𝜕𝑟 𝜕𝑟 𝑟 𝑟
1 𝜕𝑣 𝑢 𝑢 1 1 1
𝜀𝜃𝜃 = + = = (𝐶1 𝑟 + 𝐶2 ) = 𝐶1 + 𝐶2 2
𝑟 𝜕𝜃 𝑟 𝑟 𝑟 𝑟 𝑟
𝐸
𝜎𝑟𝑟 = [(1 − 𝜈)𝜀𝑟𝑟 + 𝜈(𝜀𝜃𝜃 + 𝜀𝑧𝑧 )]
(1 + 𝜈)(1 − 2𝜈)
𝐸 1 1
= [(1 − 𝜈) (𝐶1 − 𝐶2 2 ) + 𝜈 (𝐶1 + 𝐶2 2 )]
(1 + 𝜈)(1 − 2𝜈) 𝑟 𝑟
𝐸 1 𝜈 𝜈
= [𝐶1 − 𝐶2 2 − 𝐶1 𝜈 + 𝐶2 2 + 𝐶1 𝜈 + 𝐶2 2 ]
(1 + 𝜈)(1 − 2𝜈) 𝑟 𝑟 𝑟
𝐸 𝐶2
= [𝐶 − (1 − 2𝜈)]
(1 + 𝜈)(1 − 2𝜈) 1 𝑟 2
18
MECH3361/9361 Semester 2, 2016
𝐸𝐶1 1 𝐸𝐶2
=[ ]− 2[ ]
(1 + 𝜈)(1 − 2𝜈) 𝑟 (1 + 𝜈)
𝐵
=𝐴− 2
𝑟
𝐸𝐶
1 2𝐸𝐶
where 𝐴 = [(1+𝜈)(1−2𝜈)] and 𝐵 = [(1+𝜈)].
Similarly,
𝐵
𝜎𝜃𝜃 = 𝐴 +
𝑟2
(Note that since we incorporated all of the equations in our solution, it should satisfy all of
them.)
𝐵
𝜎𝑟𝑟 (𝑟 = 𝑅𝑖 ) = 𝐴 − = −𝑝𝑖
𝑅𝑖 2
𝐵
𝜎𝑟𝑟 (𝑟 = 𝑅𝑜 ) = 𝐴 − = −𝑝𝑜
𝑅𝑜 2
𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2
𝐴=
𝑅𝑜2 − 𝑅𝑖2
Therefore,
19
MECH3361/9361 Semester 2, 2016
Displacements:
𝑣=0
𝑤=0
Strains:
𝜀𝑧𝑧 = 0
Stresses:
𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2
𝜎𝑧𝑧 (𝑟) = 2𝜈 ( )
𝑅𝑜2 − 𝑅𝑖2
Fig. 4.10 Pressurised cylinder with plane strain and plane stress
𝐸(1+2𝜈) 𝜈
Replacing 𝐸 and 𝜈 by 𝐸 ∗ = (1+𝜈)2
and 𝜈 ∗ = respectively (recall Section 4.5.4), we can
1+𝜈
further obtain the solution to the corresponding plane stress problem (illustrated above).
20
MECH3361/9361 Semester 2, 2016
𝜈 𝜈 𝜈
[1 + ( )] [1 − 2 ( )] 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 1+( ) (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1
𝑢𝑝−𝑠𝑡𝑟𝑒𝑠𝑠 = 1 + 𝜈 1 + 𝜈 [ ]𝑟 + 1 + 𝜈 [ ]
[
𝐸(1 + 2𝜈)
] 𝑅𝑜2 − 𝑅𝑖2 [
𝐸 (1 + 2𝜈)
] 𝑅𝑜2 − 𝑅𝑖2 𝑟
(1 + 𝜈)2 (1 + 𝜈)2
1 + 2𝜈 1 − 𝜈 1 + 2𝜈
[ ][ ] 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 [ ] (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1
= 1 + 𝜈 1 + 𝜈 [ ]𝑟 + 1 + 𝜈 [ ]
[
𝐸 (1 + 2𝜈 )
] 𝑅𝑜2 − 𝑅𝑖2 [
𝐸(1 + 2𝜈)
] 𝑅𝑜2 − 𝑅𝑖2 𝑟
(1 + 𝜈)2 (1 + 𝜈)2
Displacements:
𝑣=0
2𝜈 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2
𝑤=− [ ]𝑧 + 𝐶
𝐸 𝑅𝑜2 − 𝑅𝑖2
Strains:
𝜈 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2
𝜀𝑧𝑧 =− [ ]
𝐸 𝑅𝑜2 − 𝑅𝑖2
Stresses:
𝜎𝑧𝑧 (𝑟) = 0
21
MECH3361/9361 Semester 2, 2016
4.8.4. Remarks
𝜎𝑟𝑟 and 𝜎𝜃𝜃 are independent of material properties. Thus, any cylinder of the same
dimensions will experience the same stress values. If strength is a major concern, one
should simply select the material with the highest strength.
However, the displacements and strains are dependent on material properties. If
stiffness is a major concern, a higher Young’s modulus material should be chosen.
When 𝑝𝑜 = 0, we have:
𝑝𝑖 𝑅𝑖2 𝑝𝑖 𝑅𝑖2 𝑅𝑜2 1
𝜎𝑟𝑟 (𝑟) = 2 − ( )
𝑅𝑜 − 𝑅𝑖2 𝑅𝑜2 − 𝑅𝑖2 𝑟 2
𝑝𝑖 𝑅𝑖2 𝑝𝑖 𝑅𝑖2 𝑅𝑜2 1
𝜎𝜃𝜃 (𝑟) = 2 + ( )
{ 𝑅𝑜 − 𝑅𝑖2 𝑅𝑜2 − 𝑅𝑖2 𝑟 2
Since 𝑟 ≤ 𝑅𝑜 , the radial stress 𝜎𝑟𝑟 ≤ 0 (always negative) and 𝜎𝜃𝜃 ≥ 0 (always
positive). Thus, 𝜎𝑟𝑟 ≤ 𝜎𝑧𝑧 (= 0) ≤ 𝜎𝜃𝜃 . Since all shear stresses are zero, the principle
stresses are:
𝜎1 = 𝜎𝜃𝜃 , 𝜎2 = 𝜎𝑧𝑧 = 0, 𝜎3 = 𝜎𝑟𝑟
y P
0 x
FEA
Theoretical solution
Fig. 4.11 Saint Venant Principles
Saint Venant observed that under pure bending, a beam conforms to a rigorous solution only
when the external forces applied at the ends of beams are distributed over the end is the same
as internal stress distribution, i.e. linear distribution.
22
MECH3361/9361 Semester 2, 2016
1. The loading area is very small relative to the overall dimensions. This means the
affected area will be much smaller than the unaffected area. For instance, in the
tensile bar in Fig 4.12, 𝐿 ≫ 𝑎, so the affected zone will be roughly |𝒛| ≈ 𝒂.
Affected Affected
unaffected zone
zone zone
a
z z
4.9.2. Examples
Tensile testing
Ultimate
stress u
Standard
Specimen Yield stress Y Fracture
Proportional pl stress f
limit
Plastic behavior
F
Elastic
behavior
When performing a tensile test of a standard material specimen, the way the specimen is
clamped should have no effect on the stress and deformation in the centre regions. Testing
23
MECH3361/9361 Semester 2, 2016
procedures require a minimum specimen length in order to avoid the end effect on the testing
result. This is an application of Saint-Venant’s principle.
Four-point bending
M(x)
Affect zone
Strain gauge testing
worse
position Better
position
For more stable and reliable testing results, strain gauges should be placed in the far field as
shown above. Strain gauges that are too close to a loading point may yield measurements that
affected by the local stress distribution.
Cantilever beams
-1000
-2000N/m
-1000 -2000
The FEA result above shows the stress distributions when end forces are applied in different
ways. From the contours, it is clear that only a small region of the beam nearest to the loading
surface is affected. Stresses in the far field are comparable in all three cases illustrated.
Example 4.4
A solid circular shaft in a machine, as shown in the figure below, is subjected to tension by a
concentrated force P, torsion by torque Q, and bending by bending moment M in the xz-plane
(i.e. 𝑀 = 𝑀𝑦𝑦 ). The radius of the shaft is R and the length is L (where 𝐿 ≫ 𝑅). Describe the
boundary conditions using Saint Venant’s principle.
x x
Q
M M
dr
d rd y
P z
P
Q
y
24
MECH3361/9361 Semester 2, 2016
Solution
Although we know the resultant force P, M and T on the ends of shaft, we do not know the
exact stress distributions that make up these resultant forces. Since 𝐿 ≫ 𝑅, Saint-Venant’s
principle means we can express these stress BCs using statically equivalent conditions (i.e.
independent of the unknown detailed stress distribution).
At 𝑧 = 0 and 𝑧 = −𝐿, for 0 ≤ 𝑟 ≤ 𝑅 and 0 ≤ 𝜃 ≤ 2𝜋 (the flat left and right end faces):
𝑑𝐴 = (𝑟 𝑑𝜃)(𝑑𝑟)
2𝜋 𝑅 2𝜋 𝑅
𝑃 = ∫ ∫ 𝜎𝑧𝑧 𝑑𝐴 = ∫ ∫ 𝜎𝑧𝑧 𝑟 𝑑𝑟 𝑑𝜃
0 0 0 0
2𝜋 𝑅 2𝜋 𝑅
2𝜋 𝑅 2𝜋 𝑅
2𝜋 𝑅 2𝜋 𝑅
𝜎𝑧𝑟 = 0
25